A qualitative model of the spatio-temporal behaviour of distributed parameter systems based on physical fields is presented. Field-based models differ from the object-based models normally used in qualitative physics by treating parameters as continuous entities instead of as attributes of discrete objects. This is especially suitable for natural physical systems, e.g. in ecology. The model is divided into a static and a dynamic part. The static model describes the distribution of each parameter as a qualitative physical field. Composite fields are constructed from intersection models of pairs of fields. The dynamic model describes processes acting on the fields, and qualitative relationships between parameters. Spatio-temporal behaviour is modelIed by interacting temporal processes, influencing single points in space, and spatial processes that gradually spread temporal processes over space. We give an example of a qualitative model of a natural physical system and discuss the ambiguities that arise during simulation.